docs: Updated Workbech section and documentation links

DorotaDR · DorotaDR · commit 6e947494a8b6 · 2026-04-20T16:52:26.000+02:00
Assisted By Cursor

Signed-off-by: Dorota Laczak &lt;dlaczak@redhat.com&gt;
diff --git a/examples/automl/churn_prediction_tutorial.md b/examples/automl/churn_prediction_tutorial.md
@@ -3,7 +3,7 @@
 
 **Scenario:** You have (or download) the **Telco Customer Churn** dataset: one row per customer, with features like contract type, tenure, charges, and a **Churn** column (Yes/No). The goal is to train a model that predicts **Churn**, so you can identify at-risk customers and use the best model from the leaderboard for retention or deployment.
 
-This tutorial walks you through that end-to-end: create a project, create S3 connections for results and training data, create a workbench with those connections attached during setup (so you do not need to restart later), add the AutoML pipeline and dataset, run AutoML with the right settings, and view the leaderboard to pick the best model.
+This tutorial walks you through that end-to-end: create a project, create S3 connections for results and training data, create a workbench with the **results** connection attached during setup (so you can reach pipeline artifacts without a restart), add the AutoML pipeline and dataset, run AutoML with the right settings, and view the leaderboard to pick the best model.
 
 ## Table of contents
 
@@ -46,7 +46,7 @@ Create two S3-compatible connections in your project: one for pipeline **results
 | **④** | Fill in the connection details: **Endpoint** (S3-compatible bucket endpoint), **Bucket** (for pipeline results and Pipeline Server artifacts), **Region**, **Access key**, **Secret key**. |
 | **⑤** | Click **Create**. |
 
-Use this connection when configuring the Pipeline Server (e.g., in **Pipeline runtimes** or project settings), so the server stores pipeline runs and artifacts in this bucket. For exact UI steps and endpoint formatting, see [Using connections](https://docs.redhat.com/en/documentation/red_hat_openshift_ai_self-managed/3.2/html/working_on_projects/using-connections_projects) and [Creating an S3 client](https://docs.redhat.com/en/documentation/red_hat_openshift_ai_self-managed/3.2/html/working_with_data_in_an_s3-compatible_object_store/creating-an-s3-client_s3) in the Red Hat OpenShift AI documentation.
+Use this connection when configuring the Pipeline Server (e.g., in **Pipeline runtimes** or project settings), so the server stores pipeline runs and artifacts in this bucket. For exact UI steps and endpoint formatting, see [Using connections](https://docs.redhat.com/en/documentation/red_hat_openshift_ai_self-managed/3.4/html/working_on_projects/using-connections_projects) and [Creating an S3 client](https://docs.redhat.com/en/documentation/red_hat_openshift_ai_self-managed/3.4/html/working_with_data_in_an_s3-compatible_object_store/creating-an-s3-client_s3) in the Red Hat OpenShift AI documentation.
 
 **Training data connection**
 
@@ -71,7 +71,7 @@ Configure the **Pipeline Server** for your project so that pipeline runs and art
 | **④** | At the bottom you will find **Advanced Settings** URI, choose **Default database on the cluster** for development or testing, or **External MySQL database** if you have an external MySQL/MariaDB for production.                                                                                                                                                                                                                                                                                                                                           |
 | **⑤** | Click **Create** (or **Save**) to create or update the pipeline server. Wait until the Pipeline Server is ready.                                                                                                                                                                                                                                                                                                                                                                                                                                             |
 
-**Note:** For more details, see [Working with data science pipelines](https://docs.redhat.com/en/documentation/red_hat_openshift_ai_self-managed/3.2/html/working_with_ai_pipelines/index) in the Red Hat OpenShift AI documentation.
+**Note:** For more details, see [Working with data science pipelines](https://docs.redhat.com/en/documentation/red_hat_openshift_ai_self-managed/3.4/html/working_with_ai_pipelines/index) in the Red Hat OpenShift AI documentation.
 
 <a id="create-workbench-with-connections-attached"></a>
 
@@ -80,8 +80,8 @@ Configure the **Pipeline Server** for your project so that pipeline runs and art
 | Step | Action                                                                                                                                                                                                                                                                                                                                                                                        |
 |------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
 | **①** | In the project, go to **Workbenches** and create a **Workbench** (notebook environment). Choose an image and resource size as needed.                                                                                                                                                                                                                                                         |
-| **②** | During workbench setup, use **Attach existing connections** to attach both the **results** S3 connection and the **training data** S3 connection you created above, so the workbench can access the buckets (e.g. to download leaderboard or artifacts later) without a restart.<br/><br/>Note: If **result** and **dataset** connections are pointing to the same bucket you will see error. |
-| **③** | Save and launch the workbench. For full steps, see [Creating a project and workbench](https://docs.redhat.com/en/documentation/red_hat_openshift_ai_self-managed/3.2/html/getting_started_with_red_hat_openshift_ai_self-managed/creating-a-workbench-select-ide_get-started) in the Red Hat OpenShift AI documentation.                                                                      |
+| **②** | During workbench setup, use **Attach existing connections** to attach the **results** S3 connection you created above, so the workbench can access the results bucket (e.g. to download the leaderboard or artifacts later) without a restart. Only the **results** connection can be attached during workbench creation; the **training data** connection is used by the pipeline via run parameters when reading data from S3, not attached here. |
+| **③** | Save and launch the workbench. For full steps, see [Creating a project and workbench](https://docs.redhat.com/en/documentation/red_hat_openshift_ai_self-managed/3.4/html/getting_started_with_red_hat_openshift_ai_self-managed/creating-a-workbench-select-ide_get-started) in the Red Hat OpenShift AI documentation.                                                                      |
 
 **Step ① — Choose workbench image and size:**
 
@@ -109,7 +109,7 @@ Configure the **Pipeline Server** for your project so that pipeline runs and art
 |------|--------|
 | **①** | Get the compiled AutoML pipeline from the repository: [pipeline.yaml](pipelines/pipeline.yaml). |
 | **②** | In Red Hat OpenShift AI, go to **Pipelines** (or **Develop & Train** → **Pipelines**) for your project. |
-| **③** | Upload the compiled pipeline as a new **Pipeline Definition** (or create a pipeline from the YAML), following [Managing AI pipelines](https://docs.redhat.com/en/documentation/red_hat_openshift_ai_self-managed/3.2/html/working_with_ai_pipelines/managing-ai-pipelines_ai-pipelines). |
+| **③** | Upload the compiled pipeline as a new **Pipeline Definition** (or create a pipeline from the YAML), following [Managing AI pipelines](https://docs.redhat.com/en/documentation/red_hat_openshift_ai_self-managed/3.4/html/working_with_ai_pipelines/managing-ai-pipelines_ai-pipelines). |
 
 **Step ③ — Upload the compiled pipeline as a new Pipeline Definition**
 
@@ -165,10 +165,10 @@ The notebook is saved under `models_artifact.path/model_name_FULL/notebooks`, wh
 | **①** | Once the AutoML run completes, check the [leaderboard](#view-the-leaderboard) to find the S3 storage path for each model's generated notebook in column "Notebook".                                                                                                                                                                                                                                                                                     |
 | **②** | **Download** the notebook to your local machine from the artifact store (S3) if you have access (e.g. via the workbench S3 connection from **Create workbench with connections attached**). The notebook is under a path like `...<run_id>/autogluon-models-training/<task_id>/models_artifact/<model_name_FULL>/notebooks/automl_predictor_notebook.ipynb` (see the [autogluon_models_training component](https://github.com/red-hat-data-services/pipelines-components/tree/rhoai-3.4/components/training/automl/autogluon_models_training) for the exact layout). |
 | **③** | Open your **workbench** (the notebook environment you created in **Create workbench with connections attached**). In JupyterLab, click the **Upload** button (upload icon) in the File Browser sidebar, select the downloaded `.ipynb` file, and upload it. The notebook appears in your workbench file tree.                                                                                                                                           |
-| **④** | Open the notebook and **run** it cell by cell. Ensure the workbench has access to the same S3 bucket (or the path configured in the notebook) so it can load the AutoGluon predictor and any data the notebook expects. If you attached the connections when creating the workbench (see **Create workbench with connections attached**), the bucket is already available.                                                                              |
+| **④** | Open the notebook and **run** it cell by cell. Ensure the workbench has access to the same S3 bucket (or the path configured in the notebook) so it can load the AutoGluon predictor and any data the notebook expects. If you attached the **results** connection when creating the workbench (see **Create workbench with connections attached**), that bucket is already available.                                                                              |
 | **⑤** | **Customize** if required: edit the model path or artifact location to point to a specific refitted model (e.g. `LightGBM_BAG_L1_FULL`), add cells for extra visualizations or metrics, change sample data, or adapt the notebook for your own workflows. Save the notebook in the workbench when done.                                                                                                                                                 |
 
-For the notebook path and artifact layout per refitted model, see the [autogluon_models_training component](https://github.com/red-hat-data-services/pipelines-components/tree/rhoai-3.4/components/training/automl/autogluon_models_training). For the overall pipeline, see the [pipeline reference](https://github.com/red-hat-data-services/pipelines-components/tree/rhoai-3.4/pipelines/training/automl/autogluon_tabular_training_pipeline). For creating and importing notebooks in the workbench, see [Creating and importing notebooks](https://docs.redhat.com/en/documentation/red_hat_openshift_ai_self-managed/3.2/html/working_in_your_data_science_ide/working_in_jupyterlab#creating-and-importing-jupyter-notebooks_ide) in the Red Hat OpenShift AI documentation.
+For the notebook path and artifact layout per refitted model, see the [autogluon_models_training component](https://github.com/red-hat-data-services/pipelines-components/tree/rhoai-3.4/components/training/automl/autogluon_models_training). For the overall pipeline, see the [pipeline reference](https://github.com/red-hat-data-services/pipelines-components/tree/rhoai-3.4/pipelines/training/automl/autogluon_tabular_training_pipeline). For creating and importing notebooks in the workbench, see [Creating and importing notebooks](https://docs.redhat.com/en/documentation/red_hat_openshift_ai_self-managed/3.4/html/working_in_your_data_science_ide/working_in_jupyterlab#creating-and-importing-jupyter-notebooks_ide) in the Red Hat OpenShift AI documentation.
 
 **Step ④ — Preview of the predictor notebook in Workbench**
 
@@ -212,7 +212,7 @@ The refit stage writes each top-N model to the pipeline workspace/artifact store
 | **④** | Enter **Model name** and optional **Description**. Enter **Version name** and set **Source model format** (e.g. custom or the format your registry uses for AutoGluon).                                                                                                                                                                                                                                                                                                                                                                          |
 | **⑤** | Click **Register**. The model appears in the Model registry and can be used for versioning, promotion, and deployment (e.g. via the single-model serving platform).                                                                                                                                                                                                                                                                                                                                                                              |
 
-For the pipeline definition and artifact layout, see the [autogluon_tabular_training_pipeline](https://github.com/red-hat-data-services/pipelines-components/blob/rhoai-3.4/pipelines/training/automl/autogluon_tabular_training_pipeline/pipeline.py) (pipeline name: `autogluon-tabular-training-pipeline`). For more on working with model registries, see [Working with model registries](https://docs.redhat.com/en/documentation/red_hat_openshift_ai_self-managed/2.22/html/working_with_model_registries/working-with-model-registries_model-registry).
+For the pipeline definition and artifact layout, see the [autogluon_tabular_training_pipeline](https://github.com/red-hat-data-services/pipelines-components/blob/rhoai-3.4/pipelines/training/automl/autogluon_tabular_training_pipeline/pipeline.py) (pipeline name: `autogluon-tabular-training-pipeline`). For more on working with model registries, see [Working with model registries](https://docs.redhat.com/en/documentation/red_hat_openshift_ai_self-managed/3.4/html/working_with_model_registries/working-with-model-registries_model-registry).
 
 <a id="prepare-the-servingruntime-for-autogluon-with-kserve"></a>
 
@@ -322,7 +322,7 @@ After the [AutoGluon ServingRuntime](#prepare-the-servingruntime-for-autogluon-w
 
 ![Model deployment - step 2](images/model_deployment_second_step.png)
 
-For more on serving and APIs, see [Deploying models on the model serving platform](https://docs.redhat.com/en/documentation/red_hat_openshift_ai_self-managed/3.2/html/deploying_models/deploying_models#deploying-models-on-the-model-serving-platform_rhoai-user).
+For more on serving and APIs, see [Deploying models on the model serving platform](https://docs.redhat.com/en/documentation/red_hat_openshift_ai_self-managed/3.4/html/deploying_models/deploying_models#deploying-models-on-the-model-serving-platform_rhoai-user).
 
 <a id="deployment-scoring"></a>
 
